Differences in Long-Term Outcome Between Patients Treated in Stroke Units and in General Wards
A 2-Year Follow-Up of Stroke Patients in Sweden
Background and Purpose— The long-term beneficial effects of stroke unit care have been proved in several randomized trials. However, there is a question of large-scale applicability in routine clinical practice of interventions used by dedicated investigators in small randomized trials. The objective of this study was to compare, 21/2 years after stroke, patients who had been treated in stroke units and those treated in general wards in routine clinical practice.
Methods— This is a prospective cohort study based on 8194 patients who were included, during the first 6 months in 1997, in Riks-Stroke, the Swedish National Register for quality assessment of acute stroke. Two years after the event, 5189 patients were still alive and 5104 were followed up with a postal questionnaire to which 4038 responded.
Results— Among the group of patients who were independent in activities of daily living (ADL) functions before the stroke, patients who were treated in stroke units were less often dependent in ADL functions, after adjustment for case mix (OR, 0.79; CI, 0.66 to 0.94). If they also lived at home before the stroke, then they had a lower case-fatality rate 2 years after the stroke (OR, 0.81; CI, 0.72 to 0.92).
Conclusions— Long-term beneficial effects of treatment in stroke units were shown for patients who were independent in ADL functions before the stroke. No benefits were shown for patients who were dependent on help for primary ADL before the stroke. Further studies on this group of patients with more detailed outcome measures are needed.
A meta-analysis of all randomized studies has shown that stroke unit (SU) care improves survival and functional outcome after stroke. The Stroke Unit Trialists’ Collaboration concluded in their latest structured review that there is a benefit of SU care and that the benefit is not restricted to any subgroup of patients or model of stroke care.1
Although a randomized, controlled study design has its advantages when studying the effect of a structured stroke care that is not confounded by other factors, there is a risk of an overly strict selection of patients.2 The patients and methods might not be representative for a true clinical picture of the disease. Furthermore, there is a question of large-scale applicability in routine clinical practice of interventions used by dedicated investigators in small, randomized trials.
An attempt to answer this question was performed earlier with data from Riks-Stroke, the Swedish National Register for quality assessment of acute stroke events.3 The analyses on early survival and functional outcome were restricted to patients who had been living at home without any community support before the stroke. A strict and uniform definition for SUs was used in Riks-Stroke, but each participating hospital could decide if they had met the criteria for an SU or not. Although the organization of the care was not identical between different hospitals in that study, all SUs were nonintensive SUs. In that study, a beneficial effect from treatment in an SU was shown, but only for the subgroup of patients who were fully conscious on arrival at hospital.3
In our previous study, follow-up was restricted to 3 months after stroke.3 In the present study, follow-up was performed 21/2 years after stroke to see if the beneficial effects of SUs are sill present in a heterogeneous stroke population that reflects the routine health care in a whole country.
Subjects and Methods
The Riks-Stroke Register
This study is a prospective cohort study based on the Swedish National Register for Quality Assessment of Acute Stroke Events, Riks-Stroke. The register started in 1994, and in 1997 almost all hospitals in Sweden treating patients with acute stroke participated in the Riks-Stroke Register. The register includes patients with hemorrhagic stroke and patients with ischemic stroke. Subarachnoid hemorrhages and transient ischemic attacks are not included in this report. Both new and recurrent strokes are included. Data collection in Riks-Stroke is kept simple to ensure maximum coverage. In short, the register provides information about age, sex, the history of previous stroke, and the life situation before the current stroke event. Items related to the acute care and to the discharge from the hospital are registered. Participating hospitals also must report whether or not the patient was treated in an organized SU. The criteria for a SU are defined in the Riks-Stroke manual.
During the first 6 months of 1997, 8194 patients had an acute stroke event that was registered in Riks-Stroke. Earlier estimation show that Riks-Stroke covers ≈80% of all stroke patients in Sweden (E.-L. Glader, MD, unpublished data, 2000).
Within the program for the Riks-Stroke register, each patient should be followed up 3 months after the stroke. The 3-month questionnaire describes life situation after discharge from the hospital in terms of living situation, dependency on help for daily activities (ADL), and satisfaction with health care after the stroke. At the 3-month follow-up, 1296 patients (16.2%) had died. A total of 6408 patients had completed the follow-up questionnaire and another 490 patients were still alive but were not followed up.
In October 1999, a second follow-up was performed on the same patients. The time since the stroke was 2 years, 3 months, to 2 years, 9 months. The average follow-up time was 30.4 months (SD, 1.7), and there were no differences between patients treated in SUs or in general wards (GWs). Of the patients who had a stroke in 1997, 3005 had died (36.7%). A total of 5189 patients were still alive, and the follow-up questionnaire was sent by mail to 5104 patients. The other 85 patients were still alive, but their current mailing addresses were not found. The patients received one reminder to answer the questionnaire. When the data collection phase was finished, 4038 had returned the questionnaire (78.8%) and 15 patients were not included in the analyses because their identification numbers had been lost. Approximately half of the patients (52.9%) had completed the questionnaire by themselves; in 22.4% it was completed by a next-of-kin alone, and in another 24.8% it was completed with the help of health care personnel or someone else.
A questionnaire was constructed for the follow-up 2 years after the stroke. It was based on the same questions as in the 3-month follow-up within the Riks-Stroke program and monitored dependency on help for primary ADL (mobility, toilet visits, and dressing/undressing) and living situation before the stroke. Additional questions concerning need of help with primary ADL (eating/drinking and personal hygiene) and secondary ADL (preparation of food, grocery shopping, and cleaning) were also included in the follow-up questionnaire.
Questions about health, which had been used in an interview study of elderly in Sweden, performed by the National Board of Health and Welfare, were included in the questionnaire.4 Items on the patient’s self-perceived depression, fatigue, anxiety, and pain and on cognitive impairments (difficulties with speaking, reading, and writing) were also included. A few questions concerned the need of support and rehabilitation. One question concerned the dependency of help and support from informal caregivers such as next-of-kin. Responses to questions about performance of hobbies or interests from before the stroke and about knowledge about whom to ask if one needed more help were also collected.
Statistical analyses were performed with the SPSS statistical package (version 10.0).5 Comparisons between patients treated in SUs and patients treated in GWs were adjusted for age difference with binary logistic regression with age as a covariable, because the patients treated at GWs were an average of 1.4 years older when they had their stroke. Multivariate binary logistic regression analyses were performed to adjust for differences in case mix at the time of the stroke event. Background variables used in case-mix adjustments were, except for age and sex, level of consciousness, marital status, examination with CT, type of stroke, number of strokes, and living condition. The multivariate models were tested in both a backward and a forward design with the same results. The survival analyses were performed with a Cox proportional hazards model. Differences were measured in odds ratios with 95% confidence intervals.
Patient Characteristics on Admission and After 3 Months
Of the 8194 patients, 5134 patients (62.7%) were treated in SUs and 2400 (29.3%) were treated in GWs. A total of 659 (8.0%) patients were treated in a setting that fulfilled neither the criteria for an SU nor a GW. These 659 patients, together with 1 patient for whom information about type of acute care was missing, were excluded from further analyses. Among patients who were independent in primary ADL functions before the stroke, 69.2% were treated in SUs, and among those dependent on help for ADL before the stroke, 63.2% were treated in SUs (OR, 1.27; CI, 1.12 to 1.24).
Patients Independent in Primary ADL Functions Before the Stroke
The mean age for the patients who were independent in ADL functions before the stroke was 74.1 years. Patients treated in SUs were an average of 2 years younger than those patients treated in GWs (73.5 versus 75.5 years, P<0.001). Table 1 shows that there were no differences before the stroke between patients treated in SUs and those treated in GWs. However, patients in the GW group more often had a cerebral hemorrhage and an impaired consciousness. After 3 months, more patients treated in GWs had died. Among those still alive, more patients treated in SUs were living at home, and they were also more often independent in ADL functions.
Patients Dependent in Primary ADL Functions Before the Stroke
Patients dependent on help for ADL before the stroke were older (≈80.2 years), and there was no difference in age between patients treated in SUs and patients treated in GWs (80.4 and 80.0 years, respectively, P=0.43). The GW patients more often had a cerebral hemorrhage and an impaired consciousness at arrival to the hospital. Three months after the stroke, more patients treated in GWs had died, but the difference was not statistically significant (Table 1).
Follow-Up at 2 Years
A total of 1721 patients, of the 5134 patients (30.2%) who had been treated in SUs for their strokes, had died at the time for the 2-year follow-up. Among patients treated in GWs, 972 of 2400 (34.0%) had died, and they had a significantly increased risk of dying within 2 years after stroke (OR, 1.23; CI, 1.15 to 1.33) as compared with patients in the SU group. Similarly, for patients that were independent in primary ADL functions before the stroke, there was a difference in case-fatality rates: 25.4% for patients treated in SUs and 29.1% for patients treated in GWs (OR, 1.18; CI, 1.06 to 1.30) (Table 2 and Figure 1). For the group of patients that lived at home without community service before the stroke, the differences in case-fatality rates were also present after adjustment for differences in case mix at the time of the stroke event (Table 3). However, this difference was not present for the group of patients that had been institutionalized before the stroke. Among patients who were dependent on ADL functions before the stroke, there was no difference in case-fatality rates between patients treated in SUs and GWs (Table 4 and Figure 2).
Patients Independent of Help for Primary ADL Before the Stroke
The mean age of the 3376 stroke patients was 74.3 years at the time they completed the questionnaire (SD, 10.6). The age difference between patients treated in SUs and patients treated in GWs was 1.4 years. More people treated in SUs were still living in their own homes 2 years after the stroke, and they were more often still independent in primary ADL functions (Table 2). After adjustment for differences in case mix, treatment in an SU was still an independent predictor for being independent of help for ADL 2 years after a stroke (Table 3). There were no differences in the proportion dependent on help for secondary ADL or the proportion living together with a next-of-kin.
Approximately one fourth of the patients had speech problems at the time of the follow-up, with no differences between patients treated in SUs or in GWs. Reading and writing were cognitive abilities that even more stroke patients had problems with, but treatment in SUs did not predict any better results than treatment in GWs (Table 2).
Two years after the stroke, the majority of the patients estimated their general health as fairly good or even very good. There were no differences between SU patients and GW patients. As for the general health, treatment in an SU during the acute phase did not predict a better outcome for other health variables such as feelings of depression, fatigue, or anxiety. However, patients treated in an SU statistically significantly less often had pain than those patients treated in a GW, also after adjustment for differences in case mix before the stroke (OR, 0.75; CI, 0.61 to 0.91).
Table 2 shows that among the patients who needed help and support, a majority were satisfied with the help, support, and rehabilitation they received, and there were no differences between patients treated in SUs and GWs. The GW patients were somewhat more dependent on their relatives than SU patients were 2 years after stroke, but the difference did not reach statistical significance. There were no differences in the proportion that could perform their old hobbies and interests between patients treated in SUs or in GWs.
Patients Dependent on Help for Primary ADL Before the Stroke
The 354 patients who required help for primary ADL had an average age of 78.5 years (SD, 9.8) at the 2-year follow-up. Patients who had been treated in SUs were older than those treated in GWs, but the age difference was not statistically significant (79.0 versus 77.6 years, P=0.19). The proportion of women had increased and was still statistically significant higher among patients treated in SUs. Table 3 shows that there were no other statistically significant differences between patients treated in SUs or GWs.
Comparisons Between Responders and Nonresponders in the 2-Year Follow-Up Study
The patients who did not answer the follow-up questionnaire after the stroke were older and more disabled, on the average, before the stroke, than the patients who answered the questionnaire (Table 5). Three months after the stroke, these differences were even more accentuated, and 1 year after follow-up (ie, in the period from 2.5 to 3.5 years after the stroke) they also had a higher case-fatality rate (OR, 2.24; CI, 1.88 to 2.67).
In earlier studies, Riks-Stroke covered ≈80% of all stroke patients in Sweden (E.-L. Glader, MD, unpublished data, 2001). Together with a response rate of ≈80% for the follow-up questionnaire, this study is unique in its coverage of stroke patients in a whole country.
In this group of patients with many old people, previous strokes, cognitive impairments, and dementia are common. Therefore, the construction of the questionnaire was focused on making both the questions and the alternatives of possible responses easily understandable. However, the patients who did not answer the follow-up questionnaire were, before the stroke, on average older and more disabled than the patients who did answer the questionnaire (Table 5). These missing data probably affected the results more in the subgroup that was ADL dependent before the stroke than for the independent group.
For the total stroke population, as well as for the group of patients that were independent in primary ADL before the stroke, we have in the present study shown an improved outcome for patients treated in SUs compared with patients treated in general wards at follow-up more than 2 years after the stroke. The subgroup of patients who were dependent on help for ADL before the stroke was much smaller, and the results are more inconclusive. The literature shows diverse results for this group of patients. In the Norwegian stroke unit trial performed by Indredavik et al,6 patients who lived in nursing homes before the stroke were excluded. Results from earlier studies that included a more heterogeneous study population suggested that the beneficial effects of treatment in SUs might be less pronounced for elderly and more severely disabled patients.7 Other studies have not shown any differences in beneficial effect of treatment in SU between subgroups, and results from these studies support the opinion that SUs should treat an unselected population of stroke patients.8,9
Several explanations are plausible for not finding any beneficial effect of treatment in an SU for patients that were functionally impaired before the stroke in our study. Because of an already existing functional impairment among these patents, the rehabilitation potential was probably smaller than in other patients. Perhaps their impairments were so severe that interventions taken place in an SU could not affect the outcome. It is also possible that our study design did not allow us to discover possible beneficial effects. More detailed measurements of improvements may be needed for this group of patients. Data collection in Riks-Stroke is kept very simple, and only few background variables are registered. Adjustments for differences in comorbidities before the stroke and other factors that influence outcome after the stroke would need to be included for these patients who have diseases and functional impairments even before the stroke.
In the subgroup of patients who were independent of help before the stroke, the differences in case-fatality rates and ADL dependence were already present 3 months after the stroke. At follow-up 2 years after the stroke, the differences were even more accentuated. However, the pattern of rehabilitation was similar for all patients, irrespective of type of acute treatment. The patients had also been equally satisfied with the amount of rehabilitation they had received. The differences in outcome at follow-up can therefore not only be ascribed to differences in care and support late after stroke but also to differences in background variables at time for the stroke and type of treatment during the acute phase. To minimize the possible influence of differences in background variable, we performed multivariate regression models, including background variables available in the Riks-Stroke register. These results confirm several other randomized controlled studies on long-term outcome after treatment in SUs.10–13
Earlier studies that focuses on differences in quality of life between patients treated in SUs and in GWs was published by Indredavik et al14 and Juby et al.15 Indredavik et al showed statistically significant differences in all aspects of the Nottingham Health Profile except for pain, and in the study performed by Juby et al, patients treated in SUs showed better psychological outcome. In contrast, our study did not show any differences in estimated health or cognitive abilities 2 years after the stroke. The heterogeneous study population and organization of the SUs at the different hospitals might have diluted any possible beneficial effects. However, patients treated in SUs estimated that they had less pain as compared with patients treated in GWs.
In conclusion, in this nationwide observational study, long-term beneficial effects of treatment in SUs applied in routine settings were shown for patients who were independent of help for ADL. No benefits were shown for patients who had been dependent on help for primary ADL before the stroke. Further studies on this group of patients with more detailed outcome measures are needed.
This National Stroke Project (Riks-Stroke) was funded by the Swedish National Board of Health and Welfare, Norrländska Stroke Fund, and the Swedish Medical Research Council (project K2000-27GX-13574-01A and K2000-27P-12314-04B to BS). We wish to thank Ann Staaf for assistance in organizing data collection for the study.
- Received April 10, 2001.
- Revision received June 14, 2001.
- Accepted July 9, 2001.
Stroke Unit Trialists’ Collaboration. Organised inpatient (stroke unit) care for stroke: Stroke Unit Trialists’ Collaboration. Cochrane Database Syst Rev. 2000; 4.
Stegmayr B, Asplund K, Hulter-Åsberg K, Norrving B, Peltonen M, Térent A, Wester PO. Stroke units in their natural habitat: can results of randomized trials be reproduced in routine clinical practice? Riks-Stroke Collaboration. Stroke. 1999; 30: 709–714.
Socialstyrelsen. På äldre dar. Socialstyrelsen, Äldreuppdraget. 2000;11.
Norusis M. SPSS for Windows. Chicago, I11: SPSS Inc; 1993.
Indredavik B, Bakke F, Solberg R, Rokseth R, Håheim LL, Holme I. Benefit of a stroke unit: a randomized controlled trial. Stroke. 1991; 22: 1026–1031.
Kalra L. Does age affect benefits of stroke unit rehabilitation? Stroke. 1994; 25: 346–351.
Fagerberg B, Claesson L, Gosman-Hedström G, Blomstrand C. Effect of acute stroke unit care integrated with care continuum versus conventional treatment: a randomized 1-year study of elderly patients: the Goteborg 70+ stroke study. Stroke. 2000; 31: 2578–2584.
Jørgensen HS, Kammersgaard LP, Houth J, Nakayama H, Raaschou HO, Larsen K, Hubbe P, Olsen TS. Who benefits from treatment and rehabilitation in a stroke unit? A community-based study. Stroke. 2000; 31: 434–439.
Indredavik B, Slørdahl SA, Bakke F, Rokseth R, Håheim LL. Stroke unit treatment: long-term effects. Stroke. 1997; 28: 1861–1866.
Rønning OM, Guldvog B. Stroke units versus general medical wards, I: 12- and 18-month survival: a randomized, controlled trial. Stroke. 1998; 29: 58–62.
Indredavik B, Bakke F, Slørdahl SA, Rokseth R, Håheim LL. Stroke unit treatment: 10-year follow-up. Stroke. 1999; 30: 1524–1527.
Jørgensen HS, Kammersgaard LP, Nakayama H, Raaschou HO, Larsen K, Hubbe P, Olsen TS. Treatment and rehabilitation on a stroke unit improves 5-year survival: a community-based study. Stroke. 1999; 30: 930–933.
Indredavik B, Bakke F, Slørdahl SA, Rokseth R, Håheim LL. Stroke unit treatment improves long-term quality of life: a randomized controlled trial. Stroke. 1998; 29: 895–899.